September 24-27, 2017, Vancouver, Canada
Workshop on: Physical Human-Robot & Human-Telerobot Interaction:
From Theory to Application for Neuro-rehabilitation
Thursday, September 28
Workshop Organizers and Technical Committee:
Call for Posters
We invite participants to take this opportunity to present their recent research on the topic of the workshop in the form of posters. One-page proposals in the standard format for the IEEE IROS conference for poster presentations should be submitted before September 4, 2017. All submissions will be judged based on relevance to the workshop topic, technical quality, and novelty. Authors of accepted abstracts are expected to attend the workshop and to present their work as posters during the interactive session. Please send your poster to firstname.lastname@example.org.
Topics of Interest:
● Safety of physical human-robot and human-telerobot interaction
● Assist-as-needed strategies in robotic/telerobotic rehabilitation
● Control of human-centered robotic/telerobotic systems for rehabilitation and assistance
● Machine learning for human-centered robotic/telerobotic rehabilitation, assistance and assessment
● Cooperative and multi-player systems for robotic/telerobotic rehabilitation and assistance
● In-home/remote robotic rehabilitation, assistance and sensorimotor assessment
8:45 - 9:00
Introductions, Opening Discussions
Neville Hogan, Massachusetts Institute of Technology, USA.
Modeling human neuro-recovery for therapeutic robotics
Dr. James Patton,
University of Illinois at Chicago, USA.
What models for learning and recovery tell us
Coffee Break #1 and Poster Session
Dr. S. Farokh Atashzar,
University of Western Ontario, Canada.
A Grasp-based Passivity Signature (GPS) map for physical human-robot interaction: Application to design of a new safety mechanism for neurorehabilitation robots
Leibniz University of Hanover, Germany.
Safe Physical Human-Robot Interaction for Next Generation of Prosthetics
Dr. Damiano Zanotto, Stevens Institute of Technology, USA.
Quantifying the effects of inertia and human-robot misalignments in lower extremity exoskeletons
Panagiotis Artemiadis, Arizona State University, USA
On the effect of walking surface stiffness on inter-limb coordination in human walking: toward bilaterally informed robotic gait rehabilitation
Lunch & Poster Presentation and Poster Session
Todd Murphey, NorthWestern University, USA.
Maxwell's Demon as a Principle of Shared Autonomy
Dr. Mahdi Tavakoli, University of Alberta, Canada.
Robotic Learning and Imitation of Physical Rehabilitation
Carlo Menon, Simon Fraser University, Canada
Force-myography, an alternative and complementary approach to electro-myography
Dr. Dana Kulic,
University of Waterloo, Canada.
Human motion measurement and analysis for rehabilitation
The incidence rate of age-related neuromuscular disorders is rapidly increasing worldwide due to an aging society. While better medical care has increased survival rates, it has resulted in even more patients in need of Neuro-rehabilitation, Motor Assessment, and Assistive (NMAA) services. This has placed a significant burden on the healthcare systems worldwide and has challenged the quality of NMAA services delivered to patients in need. The situation is particularly difficult for patients in remote areas
A potential solution is to develop smart mechatronic and robotic/telerobotic technologies that provide safe and effective means of in-hospital and in-home physical NMAA services. In this regard, robotic rehabilitation and mechatronic assistance systems have been developed and have revolutionized the field of motor therapy. Although there are several advantages with the use of these technologies, there still exist several technical, technological and control challenges among which are (a) questionable adaptability and compatibility with the sensorimotor needs of patients; (b) limited accessibility in remote areas, (c) high cost; and (d) limited features for patient-robot interaction safety. These issues are of particular concern when high control effort is needed for severely disabled patients and when the robot is to be used in a patient’s home or in remote areas under minimal monitoring.
This workshop will present recent developments in enhancing human-robot interaction in advanced intelligent robotic/telerobotic systems for movement rehabilitation and assistance. We aim to focus particularly on issues related to physical and kinesthetic interaction between patients and robotic/mechatronic systems developed to deliver NMAA services. The workshop will discuss a broad range of related subjects including safety, mathematical modeling, smart autonomy and instrumentation techniques for physical human-robot and human-telerobot interaction as well as intelligent control algorithms. All of these will be in the context of patient-centered robotic/telerobotic systems for use in rehabilitation and assistance.
One of the major research questions that will be discussed in the workshop is how to optimize the performance, effectiveness and usefulness of these systems while guaranteeing stability of physical human-robot interaction and safety of the patient. Kinesthetic interaction is the essential feature of all force-enabled robotic and telerobotic rehabilitation and systems for mechatronic assistance, and warrant special consideration in the study of safety of human-robot interaction. It should be highlighted that most rehabilitation and assistive robots are designed to generate powerful forces to deliver sufficient mechanical energy for the required motor therapy or assistance while operating near patients with disability who cannot react fast enough in case of emergency. As a result, instability in the robots, non-optimal or unsafe interaction can cause serious injury including bone, joint, and soft tissue damage. Clearly, human-robot interaction safety needs to be carefully studied, analyzed and guaranteed while avoiding conservative solutions that significantly degrade the performance and efficacy of the treatment.
Motivated by the needs mentioned above, it is important to (a) utilize effective sensory systems for the technology that monitor interactions and measure interactive modalities; (b) develop mathematical tools to analyze interaction based on the measurements; and (c) design intelligent techniques that use the developed models and can enhance patient-robot interaction to deliver optimal and safe therapeutic and assistive forces. Various aspects of these topics will be presented and discussed in this workshop.
The focus of this transdisciplinary workshop and the topics to be covered will be of interest to researchers from several disciplines. In addition to researchers who work on different aspects of rehabilitation robotics and particularly its control issues, this workshop will target researchers from several other related disciplines as explained below:
B) Bio-signal processing is at the heart of any mechatronic system designed to be used for rehabilitation and assistance. It is needed to guarantee compatible, intelligent, compliant, safe and adaptive interaction between the patient and the mechatronic system. The wide scope of signal processing applications in rehabilitation technologies together with implementation of new artificial intelligence algorithms (such as deep learning) call for novel and innovative techniques to further advance the usability and efficacy of the systems. This workshop will be useful for researchers who are interested in advanced and real-time processing of bio-signals (e.g., EEG, EMG, eye gaze, and kinematic and kinesthetic measures of body movements).
C) For people who conduct research on designing control algorithms for enhancing human-robot interaction, this workshop will introduce a very practical field in which they can find clear applications. The fact that the users of rehabilitation and assistance systems are patients who cannot react fast, who have disabilities, who may have limited range of motions, stiff/locked muscles and joints, makes this field a unique practical area of implementation for this group of researchers.
D) In addition to the above, the topics will be of interest to researchers who work on mechanical design of robots, particularly those who work on soft robotics, soft sensing, variable-impedance actuators, and compliant power transmission.
This full-day workshop deals with physical human-robot and human-telerobot interaction focusing on the relevant theory and application for neuro-rehabilitation. The workshop represents transdisciplinary research which includes not only applied sciences from the engineering sectors (such as human-robot interaction, nonlinear control systems, machine intelligence, instrumentation, compliant robots), but also medical sciences (such as neurosciences, rehabilitation sciences, human motor control and motor learning). The transdisciplinary nature of the work and the specific focus of the topic (- interaction between robots and patients during neurorehabilitation and assistance) call for an in-depth analysis of a broad range of research, existing challenges, and possible future lines of research. This will be enhanced by interactive discussions which are particularly suited for a workshop format and will be emphasized through a round-table discussion. Our main goal is to analyze different aspects and challenges facing the topic of this workshop from various perspectives. This workshop will help to highlight new directions of research as well as and shed more light on the challenges. The workshop will include tutorial presentations on different approaches to human-robot and human-telerobot interaction. The workshop will also feature poster presentations.
Abstracts of the Presentations
Title: Modeling human neuro-recovery for therapeutic robotics
Abstract: Progress in technology-enabled neuro-rehabilitation would be facilitated by a quantitative mathematical model of the recovery process. While it is appealing to assume that recovery proceeds by motor (re-)learning, that model is at best incomplete. For example, it does not account for abnormal tone and spasticity, which commonly accompany recovery. A theory of how motor behavior is recovered should at least be able to account for the main features of unimpaired motor behavior. Despite much slower actuators (muscles), communication (neural transmission) and computation (neural processing) than contemporary robots, humans exhibit remarkably superior dexterity and agility. In consequence, they also exhibit surprising limitations. For example, moving slowly and smoothly is hard for humans. Both of these observations suggest that human motor control is based on dynamic primitives, including at least three classes: submovements, oscillations and mechanical impedances. I will review evidence that stereotyped submovements are present in the earliest movements made by persons recovering after stroke (cerebral vascular accident). Moreover, the re-organization of submovements serves to quantify the progress of recovery. I will also articulate how abnormal coordinative synergies may emerge as a consequence of abnormal muscle mechanical impedance. Thus robot-enabled measurement of mechanical impedance may provide an objective measure of at least one important aspect of abnormal muscle tone.
Sun Jae Professor of Mechanical Engineering
Professor of Brain and Cognitive Sciences
Massachusetts Institute of Technology
Title: What models for learning and recovery tell us
Abstract: It has been show to be powerful to leverage what we know (i.e., existing models) about neural adaptation in neurorehabilitation. These models stem from how people learn through mistakes, and several techniques on using such models might dictate better training conditions helped by robotic technology. However, the clinical trials model has not proven to be effective test because of few samples and restrictive inclusion criteria. I will show how such "small data" problems lend themselves well to scrutiny and interrogation from modern predictive modeling and validation techniques, suggesting a more "organic" model for gathering data
University of Illinois at Chicago, USA.
S. Farokh Atashzar
Title: A Grasp-based Passivity Signature (GPS) map for physical human-robot interaction: Application to design of a new safety mechanism for neurorehabilitation robots
Analyzing the biomechanical capability of the human upper limb in absorbing physical interaction energy can play a significant role in enhancing the quality of physical interaction between humans and robots. In this regard, we have developed a new graphical representation that can quantitatively correlate the extent of the grasp pressure and the geometry of interaction to the “extent” of hand passivity. For this purpose, based on a new theoretical framework, a user study has been conducted for 11 healthy human subjects to characterize the energy absorption capability in their arm and wrist. The correlation is statistically validated. The identified user-specific Grasp-based Passivity Signature (GPS) map can be used as a graphical tool to assess the biomechanical capabilities of the user in absorbing interaction energy. In this work, the proposed GPS map is utilized in the design of a new stabilizer for force-enabled interactive systems. The controller takes into account the variation in energy absorption during haptic task execution. The goal is to optimize the fidelity of the force-field, while guaranteeing human-robot interaction stability (despite the potential existence of delays and non-passive components). The controller is termed GPS-map stabilizer. Using this architecture, if the user provides minimum to no energy absorption during a physical interaction, the controller makes a nonlinear force reflection gate tight to guarantee stability. However, when the user demonstrates high capability in absorbing interaction energy, the controller allows the forces to be reflected. The GPS-map stabilizer is an alternative for both conventional stabilizers of haptic/telerobotic systems and fixed conservative force limits in rehabilitation systems where patient-robot interaction safety is a crucial requirement.
Postdoctoral Research Associate,
Rehabilitation Engineering and Assistive Control Technologies (REACT),
and Canadian Surgical Technologies & Advanced Robotics (CSTAR),
Department of Electrical and Computer Engineering,
University of Western Ontario, London, Ontario, Canada.
Title: Quantifying the effects of inertia and human-robot misalignments in lower extremity exoskeletons
Abstract: Due to the complexity of the human musculoskeletal system and intra- inter-subject variability, robotic exoskeletons are prone to human–robot misalignments. These induce undesired interaction forces that may cause discomfort and inaccuracy in rendering desired forces. Current ergonomic designs rely on the use of auxiliary, passive DOF that mitigate human-robot misalignments but increase robot inertia which, if not actively compensated, may also generate spurious interaction forces. Using experimental data collected as healthy individuals walked in a treadmill-based exoskeleton (ALEX II), we will discuss: 1) how the introduction of controlled knee misalignments affects undesired interaction forces, and how these forces, in turn, affect kinematics and timing of the gait; 2) how the impact of misalignment compares with that of uncompensated inertia, and what interactions occur between the two factors; 3) how the effect of these factors is modulated by walking speed. Assessing the relative impact of misalignment and robot inertia on the wearer can provide useful insights on how to improve the effectiveness of future ergonomic designs, especially in those situations where the dynamics of the movement are quasi-periodic and, therefore, predictable, such as in human locomotion.
Department of Mechanical Engineering,
Stevens Institute of Technology
Title: On the effect of walking surface stiffness on inter-limb coordination in human walking: toward bilaterally informed robotic gait rehabilitation
Abstract: Robotic devices have been utilized in gait rehabilitation but have only produced moderate results when compared to conventional physiotherapy. Because bipedal walking requires neural coupling and dynamic interactions between the legs, a fundamental understanding of the sensorimotor mechanisms of inter-leg coordination during walking is needed to inform robotic interventions in gait therapy. In this work we investigate mechanisms of inter-leg coordination by utilizing novel sensory perturbations created by real-time control of floor stiffness on a split-belt treadmill. We present results of increased hip, knee, and ankle flexion, as well as increased tibialis anterior and soleus activation, in the unperturbed leg of healthy subjects that is repeatable and scalable with walking surface stiffness. We also present similar responses from two pilot studies with impaired walkers (stroke survivors). None of the existing rehabilitation approaches make use of existing mechanisms of inter-leg coordination for providing therapy. The work proposed here has the potential to transform the way robot-assisted gait rehabilitation is currently provided. The idea of evoking activity and providing therapy to the impaired leg by performing mechanical perturbations on the healthy leg, is novel, and if successful, it can revolutionize the field of robot-assisted gait rehabilitation.
Assistant Professor of Mechanical and Aerospace Engineering
Director of the Human-Oriented Robotics and Control (HORC) Lab
School for Engineering of Matter, Transport and Energy
Arizona State University
Title: Maxwell's Demon as a Principle of Shared Autonomy
Abstract: Shared autonomy--the execution of tasks involving both people and robots--is increasingly important in applications ranging from manufacturing to rehabilitation. Static and quasistatic tasks are a challenging class of shared autonomy problems, and dynamic tasks are often even more challenging because they require tight synchronization in time. This talk will focus on the Maxwell's Demon paradox and its use as a principle of shared autonomy for situations where the task is known ahead of time. The approach creates software-enabled techniques that integrate autonomy and an operator during dynamic task execution. The talk will end with case studies using these methods, including some arising from rehabilitation.
Professor of Mechanical Engineering
Northwestern University (NxR Lab: http://nxr.northwestern.edu)
Director: MS in Robotics Program http://robotics.northwestern.edu
Title: Robotic Learning and Imitation of Physical Rehabilitation
Abstract: In telerobotic rehabilitation, a hospital-based therapist and a home-based patient can sense the forces exerted by each other’s hands, thereby enabling the therapist’s (continuous) supervision over a long distance, resembling conventional hand-over-hand therapy. In this talk, we introduce methods to deliver rehabilitation training based on a Learning-from-Demonstration (LfD) framework to facilitate a therapist’s intermittent supervision (semi-supervision) of a patient for time-saving purposes. We present how such a robotic/telerobotic system can be used for movement therapies involving reaching maneuvers as well as functional tasks such that the learned motor skills directly translate to daily life. The proposed methods can encourage participation of patients in functional task training, optionally from the comfort of home, while drastically reducing the therapist’ hours of direct involvement at the same time as increasing the patient’s hours of access to therapy.
Department of Electrical and Computer Engineering
University of Alberta
Edmonton, AB, Canada
Title: Haptic communication between humans and with robots
Abstract: To understand how robots could interact with humans we investigated collaborative motor tasks between humans, using dual robotic interface that enable us to modulate the interaction between the partners. In this talk I will describe some of the surprising results elucidating involuntary coordination patterns between humans. Embodying a model of the underlying control mechanism yield a robot partner providing similar assistance benefits as a human partner.
- G Ganesh, A Takagi, R Osu, T Yoshioka, M Kawato and E Burdet (2014), Two is better than one: Physical interactions improve motor performance in humans. Scientific Reports 4: 3824, doi: 10.1038/srep03824.
- A Takagi, C Bagnato and E Burdet (2016), Facing the partner influences tit-for-tat exchanges in force. Scientific Reports 6: 35397.
- A Takagi, N Beckers and E Burdet (2016), Motion plan changes predictably in dyadic reaching. PLoS ONE 11.12: e0167314.
- A Takagi, G Ganesh, T Yoshioka, M Kawato and E Burdet (2017), Physically interacting individuals estimate their partners movement goal to enhance motor abilities. Nature Human Behavior (published on 6/3/2017).
Department of Bioengineering
Imperial College, London, UK
Title: Extraction of neural information from the electromyogram for prosthesis control
Abstract: Upper limb prostheses require man-machine interfacing to establish a link between the user intention and the commands to the robotic limb (prosthesis). This interfacing is commonly done with the remnant muscles above the amputation, either through their physiological innervation or using the surgical approach of targeted muscle reinnervation. Muscle interfacing or myoelectric control consists in the recording of electromyographic (EMG) signals for extracting control signals to command prostheses. Modern methods of myocontrol allow for the simultaneous and proportional activation of multiple degrees of freedom, based on regression approaches. These methods provide a natural control interface for the user. However, the exclusive use of EMG as a source for feed-forward control of prostheses may not be sufficient for restoring highly dexterous control. Methods that integrate the EMG information with that from other sensors, within semiautonomous systems, may potentially substantially enlarge the bandwidth for control. The talk will cover these topics with a discussion on the major challenges in filling the gap between commercial/clinical and academic methods for myocontrol.
Professor and Chair in Neurorehabilitation Engineering,
Department of Bioengineering
Imperial College London, UK.
Title: Human motion measurement and analysis for rehabilitation
Abstract: Human motion measurement and analysis is a challenging problem, due to issues such as sensor and measurement system limitations, high dimensionality, and spatial and temporal variability. Accurate and timely motion measurement and analysis enables many applications, including imitation learning for robotics, new input and interaction mechanisms for interactive environments, and automated rehabilitation monitoring and assessment. In this talk we will describe recent work in the Adaptive Systems Laboratory at the University of Waterloo developing techniques for automated human motion measurement and analysis. We will overview techniques for motion measurement, segmentation, individualized model learning and analysis, with a focus on application and clinical evaluation in rehabilitation.
Department of Electrical and Computer Engineering
University of Waterloo
Waterloo, Ontario, Canada.
Title: Force-myography, an alternative and complementary approach to electro-myography
Abstract: Myography captures data on the force produced by a muscle during contraction and has many useful applications such as controlling prostheses and exoskeletons, in rehabilitation protocols and in human-machine interfaces. Different types of myography include Ultrasound-myography (UMG), Optical-myography (OMG), Electo-myography (EMG) and Force-myography (FMG). UMG and OMG are useful for the applications described above but pose some challenges for incorporation into convenient wearable and portable devices. Development of a device for rehabilitation purposes that is wearable provides the advantage of convenient use anywhere including at patient’s home, and would have impact on reducing healthcare costs and improving patient outcomes. EMG and FMG do present strong potential for incorporation into wearable devices. EMG makes use of electrodes contacting the skin to measure the electrical activity whereas FMG measures volumetric changes in the muscle using sensors attached to bands that can be easily worn. Data on the development and use of FMG in assistive tools for neuro-rehabilitation will be discussed.
Professor, School of Engineering Science
Simon Fraser University, Burnaby, BC, Canada